Human Robot Interaction-Personality Prediction of a Human Using Humanoid Robot

Abstract

This study presents an innovative approach to predicting personality traits by utilizing Human-Robot Interaction (HRI). The research focuses on predicting personality traits based on the Big Five model. The study incorporates nonverbal cues, such as facial expressions and body language, along with verbal interaction, a 44-item Big Five Inventory (BFI) questionnaire, and expert analysis. To facilitate the interactive session and personality prediction, a humanoid robot named NAO was employed. The robot interacted verbally with the participants, and during these interactions, it captured nonverbal cues, specifically facial expressions (happy, sad, fear, angry, and surprised), head pose (looking forward, looking up, looking down, looking left, and looking right), and body poses (standing, akimbo, close arms, open arms, and thinking). For facial expression analysis, the researchers employed the Face Emotion Recognition Plus (FER+) dataset, which was trained using Convolution Neural Network (CNN). This module enabled the recognition of different facial expressions associated with emotions. The head poses module determined head angles using Euler angles, while the body pose was estimated by calculating the shoulder and elbow joint angles using the law of cosine. The proposed system was tested on 16 participants aged between 21-30 years to access traits i.e., extraversion, neuroticism, agreeableness, openness, and conscientiousness by integrating questionnaire response, human-robot interaction, and expert analysis. Results of the study indicate a significant association between the personality predictions made by the robot and the assessments conducted by psychologists. In all 16 cases, the predicted personalities were consistent with the expert opinions. This suggests that the extensive utilization of nonverbal cues, combined with verbal interaction, holds potential for personality prediction using the Big Five model. Overall, this study demonstrates an innovative approach to personality prediction, leveraging Human-Robot Interaction and integrating multiple data sources. By incorporating nonverbal cues alongside verbal interaction and expert analysis, the proposed architecture shows promise in predicting personality traits based on the Big Five model.